Lessons from the Size-Weight Illusion

Lessons from the Size-Weight Illusion

Elena Pasquinelli

Institut Nicod – EHESS

1 bis, avenue Lowendal 75007 Paris France

Abstract

The present paper focuses on the issue that the concept of perceptual illusion is not above controversy, as the study of the classical case of the Size-Weight Illusion illustrates. The extreme positions are represented by the indirect, inferential approach to perception and the direct, ecological view; the first one indicates illusions as evidence for the role of inferential processes and internal representations in perception and the second one discards the notion of illusion that goes along with the notions of inference and internal representation. Hence, the Size-Weight Illusion not only receives different explanations, depending on the specific view of perception adopted, but is also susceptible to not being classified as an illusion phenomenon at all, as on the basis of their attitude towards the Size-Weight Illusion, ecologists deny the existence of illusions in general. Nevertheless, for ecologists too, the study of the phenomena that are analogous to the Size-Weight Illusion seems to be a precious instrument of research on the quantities the perceptual systems are sensitive to.

1. Introduction

1.1 The notion of illusion

The present paper deals with the problem of the opportunity of making appeal to the concept of illusion

A discussion of the Size-Weight Illusion will introduce the terms of the debate between those who defend the theoretical role notion of illusion and those who affirm that this notion is obsolete. The Size-Weight illusion will be thus presented as a case study; the different positions that have been expressed about its origin and nature will serve to illustrate how different the approaches to illusion phenomena can be and how illusory phenomena can play a different role in different theoretical settings.

The problem of the opportunity to making appeal to the concept of illusion arises in view of the strong criticism against the notion of perceptual illusions within the frame-work of certain direct theories of perception. In particular a dichotomy will be presented between two kinds of approaches to perception: the traditional approach based on the role of internal representations and a new vague of approaches based on the role of movement.

This distinction also involves a different attention toward perceptual phenomena that are preferentially investigated and a different approach toward the notion of illusion. The traditional approach has mainly focused on the visual modality and has assigned an important place to illusions in general and visual illusions in particular. The new vague insists on the importance of studying perception in a more ecological frame-work, as a multisensory and dynamic activity; the notion of illusion tends to be discarded because of its presumed entanglement with traditional approaches.

1.2 Reasons for choosing a haptic illusion as a study case

In the classical approach to the study of illusions, visual illusions are considered the paradigm for all illusory phenomena. However, the privilege accorded to visual illusions is not mandatory, and can be arguably considered more of an artefact in the historical development of research in perception, as vision has been studied first and more intensively than other senses or than integrated, multisensory perception. It is then important to keep in mind that there exist a wide variety of perceptual illusions, such as haptic illusions.

The choice of presenting the controversy about a haptic illusion is also motivated by the importance that movement plays in the touch modality, and in particular in the sub-divisions of the touch modality that are connected with the exertion of movement and with the involvement of the muscle receptors. The SWI is in fact described as an exemplary phenomenon which concerns the active, haptic, dynamic touch and tactile-kinesthetic perception.

The term ‘haptics’ was first introduced by Revesz [Revesz, 1958] to include cutaneous and kinesthetic information. [Loomis & Lederman, 1986] refer to the haptic sensory modality in terms of ‘kinesthetic touch’: kinesthetic touch is comprised of cutaneous and kinesthetic receptors, provides information about objects and surfaces that are in contact with the subject and guides the manipulation of objects.

Active touch [Gibson, 1962, 1966] is defined as an exploratory rather than a merely receptive sense: the variations in the skin stimulation are produced by variations in the motor activity. In active touch, kinesthesia is neither to be separated nor to be simply combined with cutaneous sensations, since the patterns of change of the skin contact co-vary with the change in limb position giving rise to one and the same information about the object properties. The non-separation of the skin senses from kinesthesia is labeled ‘haptic system’.

“The sensibility of the individual to the world adjacent to his body by the use of his body will here be called the haptic system. The word haptic comes from a Greek term meaning "able to lay hold of." It operates when a man or an animal feels things with his body or its extremities. It is not just the sense of skin pressure. It is not even the sense of pressure plus the sense of kinesthesis. […] The haptic system, then, is an apparatus by which the individual gets information about both the environment and his body. He feels an object relative to his body and the body relative to an object.” [Gibson, 1966, p. 97]

The haptic system is then sub-divided by Gibson into different forms of touch, including haptic touch (when the skin and deep tissues are stimulated by the movement at the joints, as in catching an object, palpating, squeezing, etc. in order to extract information about its geometry and microstructure) and dynamic touch (when skin and joints are stimulated in association with muscular effort, as in the discrimination of weight, which is more accurate when the object is wielded, rigidity, viscosity, etc.); oriented touch (the combination of inputs from vestibular, joint and skin receptors).

Dynamic touch presently represents a rich domain of studies in the ecological direction (see for instance [Turvey, 1996]. The perception of object properties by wielding, as in the case of weight perception, is a prominent example of dynamic touch. Dynamic touch is in fact active, but it is not concerned with, for instance, finger exploration. Recently, another use of the term ‘haptics’ has appeared in the domain of computer interfaces. Computer haptics includes the technologies and processes for the generation of force-feedback stimuli to human users in virtual reality environments. The focus is again on hand exploration and manipulation.

By choosing a haptic illusion as a study case I have hence tried to follow the direction of the objections that the ecological and the sensorimotor direct approaches raise against the methodological approach of the indirect approach to perception. In fact, in their criticism against the notion of illusions, these two direct approaches to perception reproach to classic studies on illusion to forget the dynamic character of perception.

In particular, the ecological approach to perception oppose the indirect, inferential approach (to which the classical definition of perceptual illusions is due) that perceptual phenomena described as illusions can be re-described with no recourse to cognitive inference and knowledge, just by appropriately establishing the role played by movement in a theoretical characterization of perception, as well as the connections between movement and perception in the perceptual outcome.

The role of movement in the touch modality was affirmed early by [Katz, 1989. Original work published 1925].

According to Katz, movement plays a complex role in touch perception: it intensifies the action of static stimuli and prevents the habituation of the captors; movement also creates tactile phenomena in that it allows for the perception of qualities such as texture and elasticity that are not available to static touch. Finally, movement constitutes the objective pole of touch: a stimulus can be perceived both as a subjective, proximal, local sensation or as the sensation of the external, distal object which causes the experience depending on the intervention of movement, of active touch. Touch, associated with movement, thus can be considered as the sense of reality. Nevertheless, as the SWI shows, illusions can affect the haptic modality too.

On his side, the sensorimotor view of perception invokes the sense of touch as a model for the understanding of the functioning of perception in general, in virtue of the intrinsic connection between perception and movement which touch exemplifies[1].

Even in the cybernetic context, touch has been indicated as a model for active perception. In 1951-52 the cyberneticist D. Mackay had imagined an analog intelligent machine capable of actively recognizing figures and objects without necessarily possessing an internal model of the world (the possession of an internal model being considered by Mackay as a passive form of perception or reception). The mechanism on which this intelligent artifact is based is explained by the aid of an example: the actions performed by a blindfolded person. When seeking to recognize a solid triangular figure a blindfolded subject is required to move his fingers around the outline in a specific sequence. Hence, to the blindfolded person,

“the concept of triangularity is invariably related with and can be defined by the sequence of elementary responses necessary in the act of replicating the outline of the triangle.” [MacKay, 1951-1952, p. 114].

When action is involved in the constitution of a percept or in the acquisition of a concept, touch is the model and tactile exploration is the exemplary case. On the contrary, vision represents the model for passive or merely receptive perception and concept acquisition. [MacKay, 1951-52] describes the template-fitting method of recognition introduced by [Wiener, 1948] and [McCulloch & Pitts, 1943] as a passive system in which a typical pattern of the sample to be recognized is stored in the artifact as a template, an ideal model to which real triangles must be re-conducted, and indicates in visual studies the reference for this model.

2. Description of the SWI: the smaller of two objects of equal weight is judged to be heavier when lifted

One of the best known and more powerful haptic illusions is the so-called ‘Size-Weight illusion’ (SWI) or ‘Charpentier’s illusion’, since this phenomenon was first described in 1891 by Charpentier as an effect of volume on the perception of weight[2]. Briefly, the SWI consists in the fact that the smaller of two objects of equal weight is judged to be heavier when lifted. It is a robust illusion that is resilient to the observer’s prior knowledge of the actual relative weight of the objects.

Charpentier performed his experiment with two spheres of equal weight and of 40 and 100 mm of diameter respectively; the observers were allowed to look at the spheres and were asked to lift each sphere with the palm of their hand. The larger sphere was consistently reported as lighter [Charpentier, 1891]. The experiment demonstrates that the perceived weight of an object, its heaviness, does not depend only on its physical weight.

In 1894, Flournoy extended the experience to a large number of subjects and to different sorts of objects of equal mass that were to be ranked according to their perceived weight; he demonstrated that the SWI was resilient to the prior knowledge of the observer that the objects weighed the same [Flournoy, 1894]. Prior knowledge thus seemed not to influence the perception of weight, at least with active movement and blindfolded subjects (the conditions explored by Flournoy). This resilience is considered a peculiarity of illusory phenomena and is often cited in order to demonstrate the non-permeability, hence the independence, of perception from cognition.

2.1 Explanations of the SWI

A number of studies have since then followed aimed at investigating the role of mass, volume, density, gravitational cues in the perception of weight[3]. In particular, the role of movement in weight perception had been highlighted since the 19th century: [Weber, 1978. Original work published in 1934] had noticed that weight discrimination is more reliable when objects are wielded (thus, actively moved). The ability of discriminating weights of different masses by voluntary muscular exertion was termed “sense of force”, a component of the “muscular sense”[4]. The problem was then posed of the respective role of touch and of the muscular sense (which is today indicated as kinesthesis) in the evaluation of weight. The improvement in weight evaluation with active lifting seems to indicate that receptors with sensitivity for dynamic events in the muscular apparatus are involved in weight perception[5].

2.1.1 The expectation theory: the SWI is a cognitive illusion based on expected sensory feedback

Almost immediately following Charpentier’s description, the SWI was mostly explained in terms of “disappointed expectations” [Murray, et al., 1999]. Expectation theories emphasize the role of previous experience in judgments of weight: cognitive expectations based on previously acquired knowledge about the relationship between weight and volume in normal conditions (the bigger object is normally heavier than the smaller one) affect the perception of the actual weight of the object.

In connection with the expectation theories different hypotheses about the role of movement and force in the SWI have been put forward[6]. This fact leads to the identification of at least three possible variations within the expectation theories.

In the first variation, the illusion originates from the consequences of the expectation upon the characteristics of the performed movement, such as the consequent lifting force and lifting rate of the object. The motor consequences of the cognitive expectation are thus responsible for the SWI.

Following the second variation of the expectations theories, it is possible that the information about the force exerted in muscular contraction, as in the lifting of the object, arises from at least two sources: an internal neural correlate or ‘corollary discharge’ of the motor signal sent to the motoneuron pool, which is then sent to the sensory centers; and afferent discharges originating peripherally in various sensory receptors of the muscles, tendons, spindles, joints. Hence, when proving the role of movement and of the exertion of force in weight discrimination, the respective roles of sensory information generated centrally and of sensory information generated peripherally in the production of the SWI should be determined. In fact, the mismatch between the two sources of sensory information could be individuated as the proper source of the illusion.

The hypothesis of the mismatch is strongly criticized in the formulation of the third variation of the expectation theories, which proposes to restore a purely cognitive explanation of the SWI, with no recourse to erroneous motor commands and eventual corollary discharges of the motor commands.

A constant for all the variations of the expectation theories proposed is represented by the cognitive nature of the expectation. In spite of the differences between the specific mechanisms that cause the illusion, the remote cause is individuated in the existence of an explicit knowledge about the relationship between the weight and volume of objects. This knowledge creates expectations about the perceptual consequences of certain movements, such as the lifting of an object.

2.1.1.1 The cognitive-motor variant of the expectation theory

The cognitive-motor variant of the expectation theory [Ross & Gregory, 1970] affirms that the SWI is alleged to the wrong application of knowledge about objects [Gregory, 1997]. In Gregory’s view illusions are the product of a malfunctioning in perception. According to Gregory [see Gregory, 1968, 1973, 1997, 1998], two main categories of malfunctioning can be distinguished: those located at the mechanical or physical level of the sensory signals and sensory organs (optical or sensory illusions), and those that arise from the misinterpretation by the brain of sensory information (perceptual or cognitive illusions).

“Perceptions are hypotheses: illusions are misplaced hypotheses. Further, perceptual hypotheses may be misplaced, either because the (physiological) mechanisms mediating the hypothesis-generating strategies are malfunctioning; or because the (cognitive) hypothesis-generating strategies are inappropriate.” [Gregory, 1973, p. 69]

In the case of the SWI the mechanical and physical processes are not significant, but the assumptions regarding the relation of size to weight, and the inferences which are based on these assumptions, are misleading. This is why the SWI is considered by Gregory as a perceptual or cognitive illusion.

“Small objects feel heavier than larger objects of the same scale weight; muscles are set by knowledge-based expectation that the larger will be heavier, which is generally, though not always true.” [Gregory, 1997, p. 1124]

As for the mechanism which is specifically responsible for the SWI, [Ross, 1969] suggests that the illusion might be alleged to the characteristics of the lifting movement, and in particular to the force applied during the lifting of the object. As we have seen, prior experience of objects’ shapes and weights leads the observers to expect a larger object to be heavier than a smaller object. The learnt correlation between large volumes and heavy weights and the consequent expectation would hence affect the force that the observer applies when lifting the object, a bigger motor command being transmitted to the muscles involved in lifting a larger object. [Ross, 1969] used a matching procedure to investigate the SWI: subjects were asked to match via the haptic modality the weight of a visible object to that of an unseen object whose weight remained constant. As the volume of the viewed object increased its weight too had to be increased ini order to keep the heaviness of the two objects the same.

Support for the expectation hypothesis and for the role of the characteristics of the lifting movement comes for instance from a study of [Davis & Roberts, 1976] in which subjects were asked to lift in turn a large can and a small can placed on their palm, and then to report which felt heavier. The authors examined the movement profiles as the observers lifted the objects. In individuals who undergo the illusion, initial acceleration and height are reliably greater for the large object which is experienced to be lighter. Reliable differences in peak lifting acceleration or height are not observed in those few individuals who do not experience the illusion. Since it is assumed that subjects would attempt to lift all objects at the same rate, the greater velocity, acceleration and deceleration found by [Davis & Roberts, 1976] during the lift phase probably reflects the fact that observers expected the larger objects to weigh more, and therefore applied a greater lifting force, thus producing a faster lifting movement. As a consequence of the unexpected speed, the rapid adjustment in the force exerted by the muscles leads to the perception that the object weighs less than a smaller object of identical mass[7]. Lifting rate and lifting force would thus be related and could be placed at the origin of the illusion.